Label-free, non-invasive, and repeatable cell viability bioassay using dynamic full-field optical coherence...

Area of Science:

• Biotechnology
• Machine Learning in Biology
• Cell Biology

Background:

• Cell viability is crucial for disease diagnosis and treatment.
• Current cell viability assays, like trypan blue staining, have limitations in clarity and quantification.
• Label-free, non-invasive methods are needed for real-time cellular analysis.

Purpose of the Study:

• To introduce a novel, real-time method for assaying cellular viability.
• To utilize supervised machine learning with label-free intracellular dynamic activity data.
• To compare the machine learning approach with traditional trypan blue assays.

Main Methods:

• Acquisition of intracellular dynamic activity data in a label-free, non-invasive, and non-destructive manner.
• Application of four supervised machine learning models to the acquired data.
• Validation of the machine learning models against the trypan blue assay for cell death determination.

Main Results:

• The supervised machine learning models achieved a balanced accuracy of 93.92% ± 0.86% in cell death assays.
• The developed method provides clear, quantifiable assessment of cell viability.
• Demonstrated superior performance compared to the trypan blue assay.

Conclusions:

• Supervised machine learning offers a robust and accurate method for real-time cell viability assessment.
• This label-free approach overcomes the ambiguity and limitations of staining techniques.
• The method holds significant potential for applications in cytology, disease diagnosis, and treatment monitoring.